Your Digital Analytics Platform Is Lying To You. Here Is How.
When I started in digital marketing, I naturally assumed that performance will be easy to measure. Place a conversion pixel on the confirmation page and you’re ready to go! Over the years, I’ve gotten a little wiser (and perhaps a bit more cynical).
I’ve routinely seen 15%-20% discrepancies between different analytics platforms like Google Analytics, Mixpanel, Coremetrics and others. On top of that, those numbers are also almost impossible to reconcile with the popular e-commerce platforms, like Magento and Shopify. Add the performance dashboards from Facebook, your ESP and various other vendors and you routinely get 4-5X your actual revenue.
It is frustrating, to say the least. Thankfully, you don’t need to be a big data expert to understand the numbers being presented.
The Tag Game
Generally speaking, the goal of every marketing vendor out there is to cookie your traffic. Once the cookie is placed, there is usually a 30-day attribution window. If that consumer becomes a customer within that time frame, your vendor typically takes credit. For one of my clients, their email platform claimed to be responsible for roughly 50% of the 2016 revenue.
Before you dismiss this data as total bs, you have to understand what the numbers are telling you. The email platform for my client did somehow cookie 50% of the customers. Once we dug deeper we realized that most customers signed up for the newsletter or had some contact with the sales team before placing an order. In other words, if they were not on our mailing list, the chances of converting them were slim. That’s a tremendous insight!
Lies, Damned Lies, and Statistics
Now that you understand the tag game, let’s look at how the data gets interpreted and attributed. If you think that the numbers speak for themselves, consider how different news sources interpret things like job reports. The numbers can and are bent to fit a particular narrative.
As a rule, the more complex your digital marketing strategy, the more difficult it will be to untangle the attribution web. Let’s look at one typical scenario to illustrate. You have just launched a big promotion. The first email of a 3-part series went out to 80,000 people. You’ve updated your Google AdWords ads to point to the new promo landing page. You then retarget that traffic on Google Display Network and Facebook. Your promo is a success and sales are rolling in.
A week later, you decide to review the campaign performance to understand what has and has not worked. Assuming you use Google Analytics, you open your source/medium report to take a look. What you will see there is a story. Here is how to interpret it:
- This report uses last non-direct click attribution model. Direct traffic gets largely ignored and 100% of the credit goes to the last channel that was clicked.
- Thinking back to your campaign set up, you have specifically designed it to bring traffic back to your site. That means that, by definition, you have placed certain channels towards the bottom of your conversion funnel. That is likely to be your branded paid search and retargeting campaigns. Those channels will get more credit in this view.
- Google Analytics sessions are not the same thing as visits to your site. For example, a session only lasts 30 minutes, at which point a new session is created.
Even within Google Analytics, you can run a Multi-Channel Funnels report and play around with the Attribution Model Comparison Tool to get alternative stories. Both of these are beyond the scope of this article but I’m planning to cover them in the future.
What about the other channels used in your promotion? Let’s quickly recap those as well:
- Facebook Ads – the default attribution model is 28 days for clicks and 1 day for views. 100% of the credit goes to the last ad clicked. If no ads were clicked, the most recent view will take full credit. For video ads, views, and not clicks, get the credit. Facebook also includes likes and comments as click-throughs so no real click to your website has to take place. Take a look at the full details.
- Google AdWords – even within the Google universe, things are not very consistent. By default, Google AdWords uses a last AdWords click attribution model. So if someone clicked on your AdWords ad, then on a Facebook ad and then converted, Google Analytics will give credit to Facebook. On the other hand, AdWords will give full credit to itself.
- Bing Ads – Bing typically mirrors Google and, in this case, they use the same last ad click attribution model as Google AdWords.
Most of the vendors will use some sort of variation on the above examples. Since there are few standards, companies play with these settings to gain an advantage in the industry-wide tag game. The best platforms will give you an opportunity to examine the numbers from different angles instead of feeding you a single narrative.
It is not hard to protect yourself against the tag game and the white lies your analytics platform tells you. All you have to do is ask the right questions and do a little digging. Armed with the right information, you’ll be able to make a more informed decisions for your business.